The Gate line. AI infrastructure you own.
Every Innergate deployment runs inside your perimeter, on hardware you own, with no data leaving your environment. The Gate line gives you three starting points — from a single department to an organisation-wide rollout. Each one is scoped, built and installed for your workload.
Teams & single departments
GATE I
A self-contained system for a first production deployment. Sized for a department that wants its own assistant over its own documents — legal, finance, engineering — without sending anything to an external provider. Fits in an office environment: standard power, no rack, no datacentre.
Specifications
- Concurrent users
- 5–15
- GPU memory
- Up to 96 GB
- Models supported
- Up to 30B-class dense, long-context (128K+), with embeddings and reranking
- Form factor
- Tower / workstation
- Usable storage
- Up to 12 TB
Multi-department deployments
GATE II
The step up to shared infrastructure. Serves several departments at once from a single node, running larger models at production quality alongside a fleet of smaller task-specific ones. Suited to organisations consolidating AI onto one governed platform rather than letting teams adopt tools separately.
Specifications
- Concurrent users
- 30–100
- GPU memory
- 288–384 GB
- Models supported
- Large open-weight models at production quality, plus a multi-model fleet
- Form factor
- Single node
- Usable storage
- ~36 TB
Organisation-wide
GATE III
Full-scale deployment for organisations running AI as core infrastructure. Multi-node, rack-mounted, with the memory to run frontier open-weight models at full precision and serve hundreds of users concurrently. Built for regulated environments where every interaction has to be attributable and auditable.
Specifications
- Concurrent users
- 100–300+
- GPU memory
- 768 GB – 1.5 TB
- Models supported
- Frontier open-weight models at full precision, full multi-model fleet
- Form factor
- Rack, multi-node with high-speed interconnect
- Usable storage
- 76–152 TB
What's included
- Hardware, assembled and configured for your workload
- On-site installation and integration with your network and identity provider
- Model selection and deployment, tuned to your use cases
- Team training and onboarding
- Ongoing support and maintenance
Side by side
| GATE I | GATE II | GATE III | |
|---|---|---|---|
| Recommended for | Teams & single departments | Multi-department deployments | Organisation-wide |
| CAPACITY | |||
| Concurrent users | 5–15 | 30–100 | 100–300+ |
| GPU memory | Up to 96 GB | 288–384 GB | 768 GB – 1.5 TB |
| Models supported | Up to 30B-class dense, long-context (128K+), with embeddings and reranking | Large open-weight models at production quality, plus a multi-model fleet | Frontier open-weight models at full precision, full multi-model fleet |
| DEPLOYMENT | |||
| Form factor | Tower / workstation | Single node | Rack, multi-node with high-speed interconnect |
| Usable storage | Up to 12 TB | ~36 TB | 76–152 TB |
Reference configurations. Every Innergate deployment is sized to your workload, data volume and compliance requirements — final specification and pricing are scoped per project.
Hardware is supplied as a one-off purchase. Innergate software and support are contracted separately.
Not sure which fits?
Sizing depends on how many people will use the system, what they will ask of it, and the compliance requirements you work under. A 30-minute conversation is usually enough to narrow it down.
Book a 30-min call